On Dec 01 2007, at 15:40, Kris Pister wrote:
A 16 bit checksum at L4 is redundant if every L2 hop has a 32 bit MIC.
A 16 bit checksum at L4 is not sufficient protection against bit
errors without a 32 bit MIC somewhere (either L2 or L4/5, or both as
in HART).
We *could* do a form of header compression that consumes (i.e. checks)
the L4 checksum at the compressing end and reproduces it at the
decompressing end (assuming it is not hidden by ESP etc, in which case
a IPv4 style zero checksum wouldn't help either); a leaf node
decompressor would then do nothing (as the onus of checking the
checksum has been moved to the compressing end plus the MIC check).
In ROHC, we avoided this, as it runs counter to the end-to-end nature
of the checksum; also, the checksum is useful as an additional guard
against mis-reconstructed packets due to decompressor confusion.
This does not mean that this decision is the right one for 6lowpan;
this is one example of what needs to be talked about in "Problem
Statement for Stateful Header Compression".
Gruesse, Carsten
_______________________________________________
6lowpan mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/6lowpan
